Mo.py

!pip install sentencepiece
import sentencepiece as spm
import os, json, numpy as np, tensorflow as tf
from tensorflow.keras import layers, Model
import requests
from tensorflow import keras
from tensorflow.keras import layers
import tensorflow.keras.backend as K

print('1')
tf.get_logger().setLevel("ERROR")
SEED = 42
tf.random.set_seed(SEED)
np.random.seed(SEED)

# TPU 초기화
try:
    resolver = tf.distribute.cluster_resolver.TPUClusterResolver(tpu="local")
    tf.tpu.experimental.initialize_tpu_system(resolver)
    strategy = tf.distribute.TPUStrategy(resolver)
    print("✅ TPU 초기화 완료:", resolver.cluster_spec().as_dict())
    on_tpu = True

except Exception as e:
    print("⚠️ TPU 미사용, GPU/CPU로 진행:", e)
    strategy = tf.distribute.get_strategy()
    on_tpu = False

# Mixed precision
from tensorflow.keras import mixed_precision
policy = mixed_precision.Policy("mixed_bfloat16" if on_tpu else "float32")
mixed_precision.set_global_policy(policy)
print("✅ Mixed precision:", policy)

# =======================
# 1) 파일 다운로드
# =======================
def download_file(url, save_path):
    r = requests.get(url, stream=True)
    r.raise_for_status()
    with open(save_path, "wb") as f:
        for chunk in r.iter_content(8192*2):
            f.write(chunk)
    print(f"✅ {save_path} 저장됨")

DATA_PATH = "corpus.txt"
TOKENIZER_PATH = "ko_unigram.model"

if not os.path.exists(DATA_PATH):
    download_file(
        "https://huggingface.co/datasets/Yuchan5386/Prototype/resolve/main/corpus_ko.txt?download=true",
        DATA_PATH
    )

if not os.path.exists(TOKENIZER_PATH):
    download_file(
        "https://huggingface.co/Yuchan5386/Respiso/resolve/main/bpe.model?download=true",
        TOKENIZER_PATH
    )

sp = spm.SentencePieceProcessor(TOKENIZER_PATH)

pad_id = sp.piece_to_id("<pad>") if sp.piece_to_id("<pad>") != -1 else 0
start_id = sp.piece_to_id("<start>")
sep_id = sp.piece_to_id("<sep>")
end_id = sp.piece_to_id("<end>")
unk_id = sp.piece_to_id("<unk>")
vocab_size = sp.get_piece_size()
print(f"✅ Vocabulary size: {vocab_size}")

max_len = 512
batch_size = 128

def text_to_ids(text):
    return sp.encode(text, out_type=int)

def ids_to_text(ids):
    return sp.decode(ids)

def txt_stream(file_path):
    with open(file_path, "r", encoding="utf-8") as f:
        for line in f:
            text = line.strip()
            if not text:
                continue

            ids = text_to_ids(text)
            ids = ids[:max_len - 1]  # 마지막에 <end> 넣기 위해 -1

            full_input = ids + [end_id]
            pad_len = max_len - len(full_input)
            full_input += [pad_id] * pad_len

            # target = next-token shifted sequence
            target = full_input[1:] + [pad_id]
            yield (
                tf.convert_to_tensor(full_input, dtype=tf.int32),
                tf.convert_to_tensor(target, dtype=tf.int32)
            )


LIMIT = 2000000  # 원하는 만큼

dataset = tf.data.Dataset.from_generator(
    lambda: txt_stream(DATA_PATH),
    output_signature=(
        tf.TensorSpec(shape=(max_len,), dtype=tf.int32),
        tf.TensorSpec(shape=(max_len,), dtype=tf.int32),
    )
)

dataset = dataset.take(LIMIT).shuffle(2000, seed=SEED).batch(batch_size, drop_remainder=True).prefetch(tf.data.AUTOTUNE)

with strategy.scope():
    dist_dataset = strategy.experimental_distribute_dataset(dataset)

class SwiGLU(layers.Layer):
    def __init__(self, d_model, d_ff):
        super().__init__()
        self.proj = layers.Dense(d_ff)
        self.out = layers.Dense(d_model)
    def call(self, x):
        x_proj = self.proj(x)
        x_val, x_gate = tf.split(x_proj, 2, axis=-1)
        return self.out(x_val * tf.nn.silu(x_gate))

class LoU(layers.Layer):
    def __init__(self, d_model, clip_value=5.0, eps=1e-6):
        super().__init__()
        self.d_model = d_model
        self.clip_value = float(clip_value)
        self.eps = float(eps)
        self.Q = layers.Dense(d_model, dtype='float32')
        self.K = layers.Dense(d_model, dtype='float32')
        self.V = layers.Dense(d_model, dtype='float32')
        self.norm = layers.LayerNormalization(epsilon=1e-5, dtype='float32')
        self.norm1 = layers.LayerNormalization(epsilon=1e-5, dtype='float32')
        
        self.glu = SwiGLU(d_model, 320)
    def call(self, x):
        x_f32 = tf.cast(x, tf.float32)
        residual = x_f32
        x_f32 = self.norm1(x)

        q = self.Q(x_f32)
        k = self.K(x_f32)
        V = self.V(x_f32)
        g_q = (tf.nn.tanh(q) + 1.0) / 2.0
        g_k = (tf.nn.tanh(k) + 1.0) / 2.0
        score = g_q * g_k

        score = tf.cumsum(score, axis=1) # (B, L, D)
        
        # 💡 수정된 부분: 현재 토큰까지의 누적합 평균으로 정규화
        seq_len = tf.shape(score)[1]
        # [1, 2, 3, ..., L]을 D_model 차원으로 확장
        count_for_mean = tf.cast(tf.range(seq_len) + 1, score.dtype)
        count_for_mean = tf.reshape(count_for_mean, (1, seq_len, 1))
        
        # 누적합을 현재까지의 토큰 개수로 나누어 평균 누적합 계산 (B, L, D)
        score_mean = score / count_for_mean
        
        # 정규화 분모 설정
        denom = tf.maximum(score_mean, self.eps)
        score_norm = score / denom
        # -----------------------------------------------

        score_clipped = tf.clip_by_value(score_norm, -self.clip_value, self.clip_value)
        x_comb = score_clipped * V
        
        out = self.norm(x_comb + residual)
        out = self.glu(out)
        return tf.cast(out, x.dtype)

class Lo(layers.Layer):
    def __init__(self, d_model):
        super().__init__()
        self.d = layers.Dense(64, activation='silu')
        self.w = layers.Dense(d_model)
        self.norm = layers.LayerNormalization(epsilon=1e-5, dtype='float32')

    def call(self, x):
        p = self.d(x)
        p = self.w(p)
        return self.norm(p) + x

class Block(layers.Layer):
    def __init__(self, d_model):
        super().__init__()
        self.lou = LoU(d_model)
        self.lo = Lo(d_model)

    def call(self, x):
        x = self.lou(x)
        x = self.lo(x)
        return x

class CumaLM(tf.keras.Model):
    def __init__(self, vocab_size, max_seq_len, d_model, n_layers, dropout_rate=0.1):
        super().__init__()
        self.token_embedding = layers.Embedding(vocab_size, d_model)
        self.pos_embedding = layers.Embedding(max_seq_len, d_model)
        self.blocks = [Block(d_model) for _ in range(n_layers)]
        self.ln_f = layers.LayerNormalization(epsilon=1e-5, dtype="float32")

    def call(self, x, training=False):
        batch_size, seq_len = tf.shape(x)[0], tf.shape(x)[1]
        positions = tf.range(seq_len)[tf.newaxis, :]
        x = self.token_embedding(x) + self.pos_embedding(positions)
        for block in self.blocks:
            x = block(x)
        x = self.ln_f(x)
        embedding_matrix = tf.cast(self.token_embedding.embeddings, x.dtype)
        logits = tf.matmul(x, embedding_matrix, transpose_b=True)
        return tf.cast(logits, tf.float32)

def smoothed_loss_keras(y_true, y_pred, eps=0.1):
    y_true = tf.cast(y_true, tf.int32)
    mask = tf.cast(tf.not_equal(y_true, pad_id), tf.float32)
    vocab = tf.shape(y_pred)[-1]
    y_true_oh = tf.one_hot(y_true, depth=vocab, dtype=tf.float32)
    y_true_ls = (1.0 - eps) * y_true_oh + eps / tf.cast(vocab, tf.float32)
    log_probs = tf.nn.log_softmax(y_pred, axis=-1)
    per_tok = -tf.reduce_sum(y_true_ls * log_probs, axis=-1)
    per_tok = per_tok * mask
    return tf.reduce_sum(per_tok) / (tf.reduce_sum(mask) + 1e-8)

def masked_perplexity(y_true, y_pred, eps=0.1):
    y_true = tf.cast(y_true, tf.int32)
    mask = tf.cast(tf.not_equal(y_true, pad_id), tf.float32)
    vocab = tf.shape(y_pred)[-1]
    y_true_oh = tf.one_hot(y_true, depth=vocab, dtype=tf.float32)
    y_true_ls = (1.0 - eps) * y_true_oh + eps / tf.cast(vocab, tf.float32)
    log_probs = tf.nn.log_softmax(y_pred, axis=-1)
    per_tok = -tf.reduce_sum(y_true_ls * log_probs, axis=-1)
    per_tok = per_tok * mask
    mean_loss = tf.reduce_sum(per_tok) / (tf.reduce_sum(mask) + 1e-8)
    return tf.exp(mean_loss)

# =======================
# 모델 생성 & 컴파일
# =======================
with strategy.scope():
    model = CumaLM(vocab_size=vocab_size, max_seq_len=max_len, d_model=256, n_layers=1)
    dummy_input = tf.zeros((batch_size, max_len), dtype=tf.int32)
    _ = model(dummy_input, training=False)
    model.summary()

    optimizer = tf.keras.optimizers.Adam(1e-4, beta_1=0.9, beta_2=0.95, epsilon=1e-8, clipnorm=1.0)
    model.compile(optimizer=optimizer, loss=smoothed_loss_keras, metrics=[masked_perplexity])

    # 학습
    history = model.fit(dist_dataset, epochs=1, verbose=1)

model.save_weights("tf_model.weights.h5")
print("✅ 모델 가중치 저장 완료!")

def generate_text_topp(model, prompt, max_len=512, max_gen=512, p=0.9, temperature=0.8, min_len=20):
    model_input = text_to_ids(f"<start> {prompt}")
    model_input = model_input[:max_len]
    generated = list(model_input)
    for step in range(max_gen):
        if len(generated) > max_len:
            input_seq = generated[-max_len:]
        else:
            input_seq = generated
        input_padded = np.pad(input_seq, (0, max_len - len(input_seq)), constant_values=pad_id)
        input_tensor = tf.convert_to_tensor([input_padded])
        logits = model(input_tensor, training=False)
        next_token_logits = logits[0, len(input_seq) - 1].numpy()
        next_token_logits[end_id] -= 5.0
        next_token_logits[pad_id] -= 10.0
        probs = tf.nn.softmax(next_token_logits / temperature).numpy()
        sorted_indices = np.argsort(probs)[::-1]
        sorted_probs = probs[sorted_indices]
        cumulative_probs = np.cumsum(sorted_probs)
        cutoff = np.searchsorted(cumulative_probs, p)
        top_indices = sorted_indices[:cutoff + 1]
        top_probs = sorted_probs[:cutoff + 1]
        top_probs /= np.sum(top_probs)
        next_token_id = np.random.choice(top_indices, p=top_probs)
        if next_token_id == end_id and len(generated) >= min_len:
            break
        generated.append(int(next_token_id))
    return ids_to_text(generated)

print("\n\n===== 생성 결과 =====")  
print(generate_text_topp(model, "지난 2년 동안 출연연이 국가가 필요한 연구를", p=0.9))